Reconstituted high-density lipoprotein infusion modulates fatty acid metabolism in patients with type 2 diabetes mellitus.

We recently demonstrated that reconstituted high-density lipoprotein (rHDL) modulates glucose metabolism in humans via both AMP-activated protein kinase (AMPK) in muscle and by increasing plasma insulin. Given the key roles of both AMPK and insulin in fatty acid metabolism, the current study investigated the effect of rHDL infusion on fatty acid oxidation and lipolysis. Thirteen patients with type 2 diabetes received separate infusions of rHDL and placebo in a randomized, cross-over study. Fatty acid metabolism was assessed using steady-state tracer methodology, and plasma lipids were measured by mass spectrometry (lipidomics). In vitro studies were undertaken in 3T3-L1 adipocytes. rHDL infusion inhibited fasting-induced lipolysis (P = 0.03), fatty acid oxidation (P < 0.01), and circulating glycerol (P = 0.04). In vitro, HDL inhibited adipocyte lipolysis in part via activation of AMPK, providing a possible mechanistic link for the apparent reductions in lipolysis observed in vivo. In contrast, circulating NEFA increased after rHDL infusion (P < 0.01). Lipidomic analyses implicated phospholipase hydrolysis of rHDL-associated phosphatidylcholine as the cause, rather than lipolysis of endogenous fat stores. rHDL infusion inhibits fasting-induced lipolysis and oxidation in patients with type 2 diabetes, potentially through both AMPK activation in adipose tissue and elevation of plasma insulin. The phospholipid component of rHDL also has the potentially undesirable effect of increasing circulating NEFA.


Gas exchange and collection of expired air
Breath by breath VO 2 and VCO 2 were measured for 2 min every 30 min during the infusions (CosMed Gas Analyzer using Quark B 2 ; Rome, Italy). Expired air was collected into a 50 l Douglas Bag for 5 min and sampled into glass rubber-stoppered blood serum (SST) vacutainers.

C-palmitate enrichment and fl ux
Plasma palmitate tracer concentration was determined by gas chromatography (GC) (Autosystem XL, Perkin Elmer) and plasma [U- 13 C]palmitate enrichment was determined by GCcombustion isotope ratio mass spectrometry (GC-C-IRMS) as previously described ( 30 ).
Palmitate fl ux was calculated using the steady-state equation described by Steele et al. and Wolfe et al. ( 34,35 ).
where F = tracer infusion rate, p V = volume of distribution for palmitate (pre-determined at 0.04), C = fasting plasma palmitate concentration at time 1 and time 2, E = tracer (labeled palmitate) to tracee (endogenous palmitate) ratio (TTR) at time 1 and time 2, and t = time.
R d of palmitate ( mol/min/kg) is a function of R a minus the change in total plasma palmitate between two time points (t 1 and t 2 ), and as such the equation is as follows:

Palmitate oxidation rate
Palmitate oxidation was calculated by determining the amount of expired 13 CO 2 generated from the catabolism of labeled palmitate tracer. Oxidation rates are a function of the plasma enrichments, acetate correction, and VCO 2 using the equation below as previously described ( 31,36 ).
where E(CO2) = labelled CO2:unlabelled CO2 ratio in expired breath, V CO2 = volume of carbon dioxide expired per breath ( mol/min), 16 = the number of carbon atoms present in one palmitate molecule, and E = plasma tracer and a(r) = acetate correction factor.

Plasma lipid analysis
Plasma was collected and analyzed for HDL, LDL, total cholesterol, apoAI, apoB, and insulin as previously described ( 7 ). Plasma NEFA was measured using the WAKO NEFA kit (WAKO, VA) per the manufacturer's instructions. Plasma triglycerides were measured using the WAKO TrigA kit (WAKO, Japan) per the manufacturer's instructions. Plasma glycerol was measured using an EnzyChrom Glycerol Assay Kit per the manufacturer's instructions (BioAssay Systems, CA).

Lipidomic analysis (HPLC and mass spectrometry)
Before analysis, lipids were extracted from plasma (10 µl) with chloroform/methanol (2:1; 20 vol) following the addition of internal standards: 100 pmol each of ceramide 17:0 (Matreya Inc., HDL-raising agents currently in development to combat atherosclerotic cardiovascular disease could potentially reduce plasma NEFA and protect against type 2 diabetes. These postulates may contribute to a mechanistic basis for the inverse epidemiological associations between HDL and metabolic disease (21)(22)(23).
The potential interplay between HDL and NEFA is made more complex by a suite of studies showing that HDL and its receptor ABCA1 modulate pancreatic ␤ -cell insulin secretion ( 7,9,(24)(25)(26). Insulin promotes a shift away from utilization of fat as an energy substrate; thus, in the context of HDL elevation, insulin would be expected to synergize with AMPK activation to inhibit lipolysis in adipose tissue but would oppose AMPK-mediated fat oxidation in muscle. The net whole body effects of HDL elevation with regard to fatty acid metabolism and plasma NEFA concentrations are unknown and can only be investigated using an in vivo approach.
The current study aimed to determine the effects of acute HDL elevation on fat oxidation and lipolysis in patients with type 2 diabetes. The clinical relevance relates to both understanding the role of low HDL in the etiology of insulin resistance and the potential application of HDLraising therapies, including rHDL, in the treatment of insulin resistance and type 2 diabetes.

Human rHDL infusion study
The current investigation concerning fatty acid metabolism was performed in parallel with previous studies from our laboratory; thus, patient characteristics and study design have been previously reported ( 7,27,28 ). Briefl y, 13 patients with type 2 diabetes mellitus participated in the study and received 80 mg/ kg (lyophilised rHDL is reconstituted with water for injection and is infused based on protein content of the rHDL particle) (CSL Behring AG, Bern, Switzerland) over 4 h and saline placebo on separate occasions separated by at least two weeks in a doubleblind crossover study. rHDL constituents and preparation rHDL consists of apolipoprotein AI (apoAI) isolated from pooled human plasma and phosphatidylcholine (PC) from soy bean and was prepared as previously described ( 7,28 ). rHDL undergoes rapid remodeling and/or interaction with endogenous HDL ( 28 ) and has been previously shown to produce biological responses analogous to native HDL ( 29 ). The preparation did not contain any other proteins (leptin, insulin, adiponectin) likely to induce a metabolic response (data not shown).

Palmitate tracer preparation and infusion
Stably 13 C-labeled palmitate (U-13 C-palmitate) (Cambridge Isotope Laboratories) was dissolved and bound to 20% human albumin (Australian Red Cross) and delivered according to previously validated protocols (30)(31)(32)(33). NaH 13 CO 3 (1.5 mol/ kg; Cambridge Isotope Laboratories) was administered as a bolus followed by a constant palmitate infusion (0.015 mol/kg/ min) for the duration of the study. A 2 hr equilibration period was implemented to achieve steady-state tracer concentrations prior to commencement of the rHDL/placebo infusion (30)(31)(32)(33).  . Extracts were centrifuged (13,000 g , 10 min), and the supernatant was dried under nitrogen at 40°C. Lipids were redissolved in 100 µl water saturated BuOH/MeOH (1:1) containing 10 mM NH 4 COOH. Quantitation was performed by liquid chromatography electrospray ionization-tandem mass spectrometry using an Applied Biosystems 4000 QTRAP. Liquid chromatography was performed on a Zorbax C18, 1.8 µm, 50 × 2.1 mm column at 300 µl/ min using the following gradient conditions: 0-100% B over 8.0 min, 2.5 min at 100% B, a return to 0% B over 0.5 min, then 3.0 min at 0% B prior to the next injection. DAG was separated using the same solvent system with an isocratic fl ow (100 µl/min) of 85% B. Solvents A and B consisted of tetrahydrofuran: methanol:water in the ratios (30:20:50) and (75:20:5), respectively, both containing 10 mM NH 4 COOH. Quan tifi cation of individual lipid species was performed using scheduled multiplereaction monitoring (MRM) in positive ion mode ( 37,38 ). Lipid concentrations were calculated by relating the peak area of each species to the peak area of the corresponding internal standard.
Cholesterol ester (CE) species were corrected for response factors determined for each species. Total measured lipids of each class were calculated by summing the individual lipid species. NEFA was extracted using a scaled-down Dole extraction ( 39 ), followed by derivatization to its corresponding N,N,Ntrimethylethylenediamine (TMEN) iodide salt by a method similar to that described by Johnson for the trimethylaminoethylester iodide salt ( 40,41 ). Briefl y, extracted fatty acids (10 µl plasma) containing 250 pmol of FA 17:0-d3 (CDN Isotopes) were treated successively with thionylchloride (20 µl, 0.2 M in dichloromethane, 10 min RT) N,N-dimethylethyenediamine (60 µl, 10 min RT), and methyliodide (60 µl, 50% v/v in methanol, 2 min RT) with each reagent/solvent removed under a stream of nitrogen prior to the addition of the next. The TMAAfatty acids were reconstituted in 100 µl ethanol, and samples (1 µl) were injected into the Applied Biosystems 4000 QTRAP using a 200 µl/min fl ow of solvent B. Quantifi cation was performed using MRM for the loss of 59 Da corresponding to the elimination of trimethylamine. Each scan was acquired over a 1 min period, and the peak areas were normalized to the internal standard (FA 17:0 (d3)) prior to the adjustment for the relative response factor of each fatty acid.

3T3-L1 adipocyte culture and cellular lipolysis
Pre-adipocytes were cultured in normal growth media ( ␣ -MEM containing 10% serum), differentiated using the standard DMI cocktail for four days and encouraged to lipid load in the presence of 10 nM insulin. Cells were treated with HDL (50 g/ml), isoproteronol (10 M), 5-aminoimidazole-4-carboxamide ribonucleoside (AICAR) (2 mM), or phenformin (1 mM) in media containing 0.1% fatty acid-free BSA for 4 hrs. Following this, media was collected and analyzed for NEFA and glycerol release as described above. Cells were harvested and lysates were subjected to Western blot analysis.

Western blotting
Cells were harvested and the phosphorylation of signaling molecules AMP-activated protein kinase (AMPK) and the key downstream modulator of fatty acid ␤ oxidation, acetyl-CoA carboxylase (ACC), were determined by Western blot as previously described ( 7 ).

Statistics
Normally distributed data were compared by t -tests or repeated measures ANOVA with least signifi cant difference post hoc tests used to compare individual means as appropriate. The order of the rHDL and placebo infusions was included as a between-subjects variable in the analysis of the clinical studies. Non-normally distributed data were compared by Mann-Whitney Rank Sum tests or Kruscal-Wallis one-way ANOVA on ranks with Dunn's post hoc tests to compare individual means as appropriate. Results are expressed as means ± SEM unless otherwise indicated. All analyses were conducted using SPSS (version 16). Cell culture data represent a minimum of three separate experiments performed in triplicate. P < 0.05 was considered signifi cant.

RESULTS
The characteristics and demographics of the patients in these studies have been detailed previously ( 7,27,28 ).
Consistent with increased fat oxidation, plasma glycerol signifi cantly increased throughout the placebo infusion; however, this was completely inhibited during rHDL infusion (% change at end of infusion: placebo = 43 ± 19.5%,

Lipidomics analysis
The concentration of individual families of lipid species before (pre) and after (post) both infusions for each group, including DAG, ceramide (Cer), free cholesterol, CE, PC, LPC, and NEFA, are shown in Fig. 4 . All were unchanged after the placebo infusion, and there was no effect of rHDL on total DAG or Cer. Small, but signifi cant increases were seen in free cholesterol and cholesterol esters after rHDL infusion. There were signifi cant increases in total PC (1563 ± 96 µmol/l versus 2944 ± 117 µmol/l for placebo and rHDL, respectively); LPC (221 ± 19 µmol/l versus 740 ± 37 µmol/l for placebo and rHDL, respectively); and NEFA (833 ± 49 and 1218 ± 115 M for placebo and rHDL, respectively) following 4 h of rHDL infusion ( Fig. 4 ), consistent with the composition of the rHDL infusate.

DISCUSSION
The main fi nding from the present study was that an acute (4 hr) infusion of rHDL inhibited fasting-induced adipose tissue lipolysis and fat oxidation in humans with type 2 diabetes. Inhibition of lipolysis possibly results from the dual effects of HDL-mediated insulin release ( 7 ) as well as activation of AMPK in adipose tissue. While insulin and AMPK act synergistically to reduce lipolysis in adipose tissue, it was not possible to determine which mechanism predominated after rHDL infusion. However, studies in adipose cell culture suggest for the fi rst time that HDL increases AMPK signaling and inhibits lipolysis. The observed reduction in whole body fat oxidation after rHDL indicates a preponderance of an insulin effect over AMPK activation in skeletal muscle. Secondarily, this study also suggests that plasma enzyme-mediated turnover of PC moieties in the rHDL infusate leads to signifi cant elevations in plasma NEFA.
Consistent with prolonged fasting, we observed a reduced plasma insulin level, together with increased rates of palmitate release (R a ), uptake (R d ), and oxidation in the placebo trial, indicative of increased adipose lipolysis and utilization of fatty acids as a substrate for ATP production. rHDL prevented this fasting-induced increase in lipolysis observed during the placebo infusion, as demonstrated by stable palmitate R a , R d , and oxidation rate. Because these data suggested that rHDL was potentially inhibiting fasting-induced lipolysis, two well-described plasma markers of lipolysis ( 42 ), NEFA and glycerol, were measured. Consistent with palmitate tracer data, the increases in plasma glycerol concentration observed during the placebo trial were abolished during the rHDL infusion. In this context, rHDL would also be expected to blunt fasting-induced elevations in NEFA; however, this was not observed, and in fact, NEFA was signifi cantly increased beyond the placebo level by rHDL. Adipocyte cell culture studies were conducted to resolve this disparity and establish the direct effects of HDL on adipocyte lipolysis.
On the basis of previous data from our group and others, we hypothesized that HDL-mediated AMPK activation in adipocytes would inhibit lipolysis. This was confi rmed with the demonstration that HDL both increased the phosphorylation of the AMPK target ACC and reduced NEFA and glycerol release from adipocytes. This inhibition was of a similar magnitude to that observed with AICAR and phenformin, two well-known activators of AMPK. While AMPK phosphorylation was not signifi cantly increased by HDL, this fi nding is consistent with the previously welldescribed phenomenon that AMPK is generally transiently activated, with consequent sustained stimulation of downstream targets, such as ACC ( 7,43,44 ). The fi nding that AMPK inhibits lipolysis in adipocytes has been demonstrated in a number of previous studies in vitro ( 17,45 ) and in vivo in rats ( 15 ). A recent study also demonstrated that systemic administration of AICAR in humans reduced whole body lipolysis in a manner similar to that observed in the current clinical trial ( 16 ). Given the similarities in the data from the study by Boon et al. and our current data, we believe this provides support in favor of rHDL acting through an AMPK-related mechanism in adipose tissue. To our knowledge, this is the fi rst report to demonstrate that HDL can activate AMPK in adipocytes and inhibit hibitory effect of HDL on lipolysis in vitro to elevate NEFA during rHDL infusion. NEFA would be derived from PC via the action of enzymes in the phospholipase A2 group, resulting in production of various NEFA and LPC species ( 46 ). While this analysis provides a snapshot for the circulating concentrations of various lipids, we acknowledge that it cannot characterize the complexity of lipid interactions occurring during the 4 h infusion period. These data strongly imply that the increased circulating NEFA observed after rHDL infusion is due to remodeling of the rHDL infusate and not due to direct signaling events elicited by rHDL. This fi nding thus allows us to more confidently interpret the accompanying data demonstrating that rHDL is indeed inhibiting lipolysis.
Taken together, the current data indicate that an acute infusion of rHDL signifi cantly alters fasting-induced fatty acid turnover and oxidation. This is likely driven, at least in part, by inhibition of adipose tissue lipolysis as a result of activation of AMPK and HDL-induced increases in insulin secretion. These observations are consistent with the lipolysis. This direct effect would be reinforced by the additional actions of HDL on pancreatic insulin secretion ( 7 ).
The inconsistency between HDL-mediated inhibition of lipolysis in cell culture and the elevated NEFA after rHDL in the clinical trial implicated the rHDL infusate as the source of NEFA elevation. The infusate was prepared, as described previously ( 7 ), by combining human apoAI with soybean PC. This preparation undergoes rapid remodeling upon infusion ( 28 ), and we hypothesized that modifi cations to the infused PC would result in liberation of NEFA. LPC species compose only a small fraction of total phospholipids in the rHDL preparation, yet plasma concentrations of LPC were shown to rise substantially in conjunction with similar NEFA subspecies. Conversely, considering the high levels of PC species such as 34:6 in the rHDL infusate, these species did not rise in plasma to the levels that would have been expected after rHDL infusion. These data strongly imply that the actions of the phospholipase A2 super family on infused PC overrides the demonstrated in- rHDL, reconstituted HDL. a 4 h values that were signifi cantly different between rHDL and placebo infusions ( P < 0.05). b Species that made the majority contribution to the increase in plasma concentration seen after rHDL infusion, as indicated in "Results." hypothesis that low plasma levels of HDL observed in patients with type 2 diabetes may contribute to increased circulating NEFA and the pathophysiology of type 2 diabetes. However, while this study highlights the capacity for rHDL to modulate fatty acid metabolism in adipocytes, the phospholipid component of the rHDL infusate did directly increase circulating NEFA. Our fi ndings suggest that HDL elevating agents not associated with PC infusion, such as CETP inhibitors or apoAI transcriptional upregulators, may also inhibit adipocyte lipolysis. Detailed investigation of fatty acid metabolism is therefore warranted in future investigations of chronic HDLelevating agents.